Snapmirror large snapshots issue

[BOOSTMR2VA]

Hello everyone, I am new to the community and still somewhat new to NetApp and learning fast.

I am having one issue that I need some clarification on.  We run snapmirrors of several volumes to a DR site.  Some of these snapmirrors are behind several days, and it appears the snapshots associated with said snapmirrors have become quite large (As I understand it, netapp takes snapshots for the snapmirror process).  I have already adjusted the volume sizes to accomadate the extra snapshot overflow.

My question is, once the snapmirror sync is complete, do these snapshots hang around, or immediately go away?  I want to resize the volumes back to their original sizes once the snapmirrors are complete.

On another note, I notice some voumes that have no configured snapshots, just their snapmirror to the DR site, and have quiesced snapshots on them.  Are these OK to remove?  Do they get automatically consolidated to more recent snapshots when I remove them?

Thank you in advance for any information on this.  I have searched technical documents but just need a little clarification.

[chute · Accepted solution]

My question is, once the snapmirror sync is complete, do these snapshots hang around, or immediately go away?

The snapshots that are growing are the changes that have been made since the last successful update. Once the current transfer is complete the "extra" snapshots will start transferring. Then changes made during that update will be queued into another snapshot and so on and so on. You may need to adjust your snap reserve to account for the rate of change. Or, if the bandwidth that is provided is not pushing updates quick enough, look into increasing bandwidth. If you are pushing your changes fast enough across the pipe to keep up with the rate of change on the source, then yes, these "extra" snapshots go away. Only the baseline is left.

On another note, I notice some voumes that have no configured snapshots, just their snapmirror to the DR site, and have quiesced snapshots on them.  Are these OK to remove?  Do they get automatically consolidated to more recent snapshots when I remove them?

I think perhaps a different question would be, why are the quiesced in the first place? If they are needed for DR and are quiesced, then they are not in sync. If the DR relationships are needed, then correct the relationships and ensure they are transferring. If the relationships are NOT needed. Then do the right thing and clean up the relationships. First, obtain customer approval that the relationships are not needed. Then gracefully break, delete snapshots, and remove schedules from snapmirror.conf. Last thing you want is for a customer to come back and say "hey what happened to my DR stuff" so obtaining customer approval before proceeding is highly recommended.

[BOOSTMR2VA]

Very good info, thank you!  Just to clarify, Once the snapshot is replicated, it should go away and space should be reclaimed, is this correct?

Also, the quiesced snapshots I found were not noted as incomplete, and infact had very old dates tied to them.  The snapmirror is in an idle, snapmirrored state, so it appears they are from an old issue where perhaps the snapmirror relationship broke in the past. 

[radek_kubka]

Hi,

If these snapshots are not marked as busy, they are quite likely some left-overs & OK to delete (providing you do not treat them as backups).

Can you, just in case, post output of snap list vol_name ?

Regards,
Radek

[BOOSTMR2VA]

I actually already deleted them.  We do not keep backup snapshots of this particular volume I found them on, so i removed them.  I see no impact so far.

[chute]

Just to clarify, Once the snapshot is replicated, it should go away and space should be reclaimed, is this correct?

Yes, the baseline snapshots will still exist, but the incremental will go away and you will have the space reclaimed. Keep in mind, the incremental snapshots may come and go depending on how many changes take place on the source. So just be sure your snap reserve can support the rate of change.

[BOOSTMR2VA]

I'm using Oncommand to manage the issue right now.  I have 340 GB replicated of a 517GB total snapshot so far (which is still growing as time goes, hopefully not faster than my replication speed).  Going to take all night to finish, so we will see how this goes.  I am marking this as answered, thank you for the help guys.